30
Numerical and Physical Experiments of Wave Focusing in Short- Crested Seas Félicien Bonnefoy, Pierre Roux de Reilhac, David Le Touzé and Pierre Ferrant Ecole Centrale de Nantes, France Rogue Waves’2004, Brest

Numerical and Physical Experiments of Wave Focusing in ... · (Bonnefoy et al ISOPE’02, Bonnefoy et al OMAE’04) rekmaev wadetnmege s: 3D•In • Improved control laws (Dalrymple

  • Upload
    others

  • View
    6

  • Download
    0

Embed Size (px)

Citation preview

  • Numerical and Physical Experiments of Wave Focusingin Short- Crested Seas

    Félicien Bonnefoy, Pierre Roux de Reilhac, David Le Touzéand Pierre Ferrant

    Ecole Centrale de Nantes, France

    Rogue Waves’2004, Brest

  • Topic of the talk

    Generation of deterministic wavepackets in a numerical or physicalwave basin

    •Wavemaker motion

    •Fully-nonlinear waves

    •In 2 and 3 dimensions

    (mono- and multidirectional)

    Numerical tool:

    •Spectral method

    •with high-order technique

    •Non-periodic

    •Specific treatment for wavemaking

    Wave elevation record

    ( )tη

    Time in s.

  • Theoretical Framework

    • Potential flow theory

    and

    • Free surface potential

    • Fully-nonlinear free surface conditions

    on

  • unknown: the only nonsurfacic quantity

    Time-marchingRunge-Kutta 4

    Nonlinear Free surface equations

    Time evolution strategy

    on

    on

    Separately approximated by an High-Order technique

  • Standard Higher-Order Techniques

    • The two main methods available are:

    – Higher-Order Spectral HOS(West et al 1987, Dommermuth and Yue 1987)Formal and Taylor series expansion of the potential only (not for equations)to obtain the vertical velocity.

    – Dirichlet to Neumann Operator DNO(Craig and Sulem 1993)Formal and Taylor series expansion of the DNO only (not for equations) iethe normal velocity.

    • Decomposition in recursive Dirichlet problems solved byFourier spectral method and collocation nodes

  • High-Order Method

    • Advantages:– Fast solvers with computational costs in O(NlogN) thanks to the use

    of Fast Fourier Transforms.Large number of wave components for random seas or steep wave fields.

    – High accuracy of the spectral methods

    • Limitations of the HOS method:– Non-breaking cases

    Steep wave field involves high order nonlinearities– increase the number of modes– dealiase carefully

    – Sawtooth instabilities for very steep wave calculations Standard five-point smoothing applied regularly through the

    steepest simulations or decrease of the number of modes

    • Standard Higher-Order Simulations– Periodic boundary conditions on the free surface– Initial stage: Free surface elevation and potential specified at t=0– Pneumatic wave generation

    F(x+Lx , y) = F(x , y)F(x , y+Ly) = F(x , y)

  • A High-Order Approach for Wave Basin

    • Basin with rigid wallsBy simply changing the basis functions on which we expand our solution:

    The natural modes of the basin

    • Cosine functionsStill possible to use Fast Fourier Transforms

    • A wavemaker to generate the waves starting from rest(no initial wave description required)

    • The concept of additionnal potential (Agnon and Bingham 1999)

    •Inlet flux conditionsolved in an extended basin

  • • Resolution of the additional potential in a extended basin

    • Extensively validated in a previous second order model(Bonnefoy et al ISOPE’02, Bonnefoy et al OMAE’04)

    • In 3D: segmented wavemaker• Improved control laws

    (Dalrymple method for large wave angles)

    Wavemaker modelling Inlet Flux Condition

    Also solved by spectral method

  • Applications

    • Improved deterministic reproduction technique in 2D

    • Deterministic reproduction of directional focused wavepackets in 3D

    Wave elevation record

    ( )tη

    Time in s.20 30 40

    -0.2-0.15

    -0.1-0.05

    00.05

    0.10.15

    0.2

    Time in s.

    Wavemaker motion

    ( )X t

  • Deterministic reproduction in 2D

    • Wavemaker motion to reproduce this wave field– Control in the frequency domain with a set of components:

    amplitude, phase (+ angle in case of 3D generation)

    ������

    ���

    ������

    ������

    �����������������������������������������������������������������������������������������������������������������������������������������

    �����������������������������������������������������������������������������������������������������������������������������������

    ���������������������������������������������������������������������������������������������������������������������������������������������

    ��

    z

    x

    ��

    ��

    ��

    X(t)

    Wave probe

    X̂(f) C∈

    Wavemaker

    ( )X t

    Wave elevation record

    ( )tη

    Time in s.

    Characteristics:•Steep wave packet: kpAl = 0.26•Asymetric in time

    Basin dimensions: 50m long 5m deep

  • Analytical methods

    • Linear backward propagation: reverse phase method(e.g. Mansard and Funke 1982)

    • Second-order bound correction of amplitudes(e.g. Duncan and Drake 1995)

    ������

    ���

    ����

    ����

    ����

    ��������������������������������������������������������������������������������������

    ��������������������������������������������������������������������������������������

    ��������������������������������������������������������������������������������������������

    ��

    z

    x

    ��

    ��

    �X(t)

    Wave probe

    Wavemaker

    ( ) ( ) ( ) ( ) ( ) ( )ˆˆ ˆ ˆikx ikxt f f e X f TF f e X tη η η η→ → → = →-1F F

    ( ) ( ) ( ) ( )1 2ˆ ˆ ˆt f f fη η η η→ = +F

    Wave elevation record

    ( )tη

    Time in s.

    20 30 40

    -0.2-0.15

    -0.1-0.05

    00.05

    0.10.15

    0.2

    Time in s.

    Wavemaker motion

    ( )X t

    Wavemaker Transfer Function

  • Iterative corrections

    TargetBefore iterationAfter 5 iterations

    Initialguess

    Non-linearsimulation

    Correction on amplitudesand/or phases

    Correctedmotion

    n=1

    ηn (t)Comparison tothe target afterFourier analysis

    ( )expn n nX X i φ=

    ( )nX f

    Target wave

    $ ( )fη

    Time in s.

    TargetFirst order inputSecond order input

    Time in s.

    Without iteration With iterations

    Elev

    ati

    on in m

    .

    Elev

    ati

    on in m

    .

  • A first step towards higher ordercontrol of nonlinearities

    • In litterature: analytical-empirical approach Clauss et al (OMAE’04)

    • Crest and trough focusing

    • Johannessen et Swan (PRSL 2001)• Zang et al (OMAE 2004)• Bateman (PhD Thesis 2001)

    • Separation in odd and even orders

    • Phase modification by third order effects is present in odd and evenelevation

    ( )tη ( )tπη

    ( )

    ( )

    1212

    odd

    even

    π

    π

    η η η

    η η η

    = − = +

    ( ) ( ) ( )( )

    ( ) ( )( )

    ' ' ''

    ' '

    ' m n p m n pn n

    m n m n

    i t k k k xi t k xodd n m n p

    n m n p

    i t k k xeven m n mn

    m n

    a e a a a e

    a a G e

    ω ω ωω

    ω ω

    η

    η

    ± ± − ± ±− ± ±

    > ≥

    ± − ±± ±

    ≈ + ≈

    ∑ ∑

  • Validation with a small amplitude wave packetSecond order effects

    • 10 cm amplitude wave packet (at the focusing point) for 5 m meanwavelength

    • Nonlinear effects reduced to second order

    • Good agreement between first order and odd elevation,and between second order and even elevation

    First orderOdd elevationSecond orderEven elevation

    Target wave packetCrest focusingTrough focusing

    Measured elevations Odd-even decomposition

    Time in s.Time in s.

  • Third order effects for higher wave amplitude

    • Resonant Interactions• No instabilities detected (in contrast with Johannessen and Swan (PRSL 2001)

    • Phase velocity

    • Non resonant Interactions• Bound terms

    • Example with a 30 cm wave packet

    Odd and linear elevation Even and second order elevation

    linear phase velocitynonlinear phase velocityeven elevation

    linear phase velocitynonlinear phase velocityodd elevation

    Time in s.Time in s.

    Elev

    ati

    on in m

    .

    Elev

    ati

    on in m

    .

    ( )'n ni t k xodd n

    na e ωη −=∑

    To build the linear elevation

  • Application to deterministic reproduction

    Initial decomposition : linear Initial decomposition : second order

    The main features of the focused target wave packet are well reproduced withonly one correction of the wavemaker motion (no iteration so far)

    •Central crest and lateral troughs are close to the target both in amplitudes and phases

    •Central crest amplitude is correctly estimated

    • Better control of the high-frequency waves

    Time in s. Time in s.

    Elev

    ati

    on in m

    .

    Elev

    ati

    on in m

    .

    Wavemaker motion corrected with the phase shift

    due to nonlinear phase speed modification( ) ( )' iX f X f e φ∆=

  • Focused wave packet reproduction in 3D

    • Directional irregular wave field

    S(f,θ) = S(f) D(θ,f)

    • Modified Pierson-Moskowitz spectrum (fpeak=0.5Hz, Hs= 4 cm)

    • Directional spreading with s=10

    • Focusing time t=45 s

    • Elevation recorded in 5 locations(probe array used for short-crested seas analysis)

    t = 25 s. t = 45 s.

    ( )5 425 5exp

    4 4Sp ps

    p

    f fHff f f

    = −

    ( ) 2, cos2

    D s of θ θθ − =

  • Reproduction of a directional focused wave field

    Analysis in the frequency domain (for the 5 probes)

    ( )( cos sin ) for 1 to 5i ik x y pa e e FT pφ θ θ η− + = =Three unknowns at each frequency : , ,a φ θ

    A set of nonlinear equations solved with a nonlinear least squares method(local minima are expected)

    and different initial guesses

    1 to 5

    random between 0 and 2 gaussian random angle

    init p

    init

    init

    a FT

    φ πθ

    = =

    ( ) ( ) ( )25

    01

    T targetp p

    pw t t t dtε η η

    =

    = −∑∫

    We obtain a set of solutions of the nonlinear equations: we choose the onethat minimises

  • Directional focused wave field

    Simulated wave packets with the HOS model of the wave basinfor both the focused target and the reproduced wave packet

    Target wave fieldfp = 0.5 Hz, Hs = 4 cm

    Directional spreading s=10

    Focusing time t = 45 s

    Reproduced wave field

    ( ),, , y ta Xφ θ →Prescribed snake-like wavemaker motion

    Large waves angles generated with the Dalrymple method

  • Directional focused wave field

    View of the wave field before the focusing event at t = 33.5 s

    Target wave field Reproduced wave field

    •The main features of the focusing packet are correctly reproduced

    •The high-fequency range is underestimated in the predictedwavemaker motion

  • Directional focused wave field

    View of the wave field at the focusing event t = 45 s

    Target wave field Reproduced wave field

    •Underestimation of the wave crest

    •Overestimation of the width of the crest

  • Conclusion

    • High-Order Spectral method applied to the wave generation in awave basin

    • Improvement of the wavemaker motion for the generation of deterministic wave packets

    • Part of third order effects (phase velocity) taken into account in 2D

    • Attempt of deterministic reproduction in 3D

    Future work

    • Phase velocity correction applied iteratively• Application to different kinds of wave packets (narrow-

    banded, broad-banded…)

  • Comparison between numerical simulations andexperiments

    Amplitude 40 cm Amplitude 30 cm

  • Nonlinear simulation results for linear and second-order wavemaker motion

    Difference with the target signal:– Time shift– Peak amplitude– Through amplitude

    Improvement of the second ordermodel:

    – Reduced time shift

    Basin dimensions: 50m long5m deep

    Number of modes: Nx=512Nz=64

    Order of decomposition: M=5

    TargetFirst order inputSecond order input

    Time in s.

    Analytical methods

    Elevation in m.

  • Wavemaker control

    • Successive corrections

    • Initial guess: second-order analytical model

    Initialguess

    Non-linearsimulation

    Correction on amplitudesand/or phases

    Correctedmotion

    n=1

    ηn (t)Comparison tothe target afterFourier analysis

    ( )expn n nX X iφ=

    ( )nX f

    Target wave

    $ ( )fη

  • Numerical checkings

    • Number of modes

    • Order of decomposition for

    • Time interval between successive smoothing (dt = 0.014s)

    reference0.45201024

    0.08 %0.4516512

    2.1 %0.4459256

    Relative errorPeak heightNumber of modes

    zφ∂∂

    reference0. 45165

    0.18 %0.45244

    0.40 %0.45343

    Relative errorPeak heightOrder M

    0.09 %0.451930

    0.15 %0. 451620

    0.02 %0.452240

    Reference0.452360

    Relative errorPeak heightNumber of timesteps

  • Iterative corrections1-Phase lag refinement

    Wave elevation obtained at iteration « n »

    with motion

    Correction on the phaseφn+1 = φn + ∆φ with ∆φ = φ − ψn

    Simulation result after 5 iterationsTargetBefore iterationAfter 5 iterations

    ( )expn n nX X i φ=( )exp nn n iη η ψ=

    Time in s.

    Elevation in m.

    Target phase

  • Iterative corrections2-Amplitude and phase

    Correction of the phaseφn+1 = φn + ∆φ with ∆φ = φ − ψn

    Correction of the amplitude

    with ∆a =

    Simulation result after 5 iterationsTargetBefore iterationAfter 5 iterations

    1n nX X TF a+ = + ∆

    Target amplitude

    Elevation in m.

    Time in s.

    ( ) ( )nf fη η−

  • Iterative Corrections

    14163537Error (%)

    Iterations onboth the phaseandamplitudes

    Iterations onthe phase

    First plussecond order

    First orderCase

    Wave elevation record

    ( )tη

    Time in s.

    Comparisoninterval

    2

    1

    2

    1

    target simu

    target

    100

    T

    TT

    T

    dtError

    dt

    η η

    η

    −=

    ∫∫

    Elevation in m.

    Time in s.

    •Discrepancies for the peak amplitude

    •Spurious oscillations before the focused event

  • Numerical and Physical Experiments of Wave Focusing in Short- Crested SeasTopic of the talkTheoretical FrameworkTime evolution strategyStandard Higher-Order TechniquesHigh-Order MethodA High-Order Approach for Wave BasinWavemaker modelling Inlet Flux ConditionApplicationsDeterministic reproduction in 2DAnalytical methodsIterative correctionsA first step towards higher order control of nonlinearitiesValidation with a small amplitude wave packetSecond order effectsThird order effects for higher wave amplitudeApplication to deterministic reproductionFocused wave packet reproduction in 3DReproduction of a directional focused wave fieldDirectional focused wave fieldDirectional focused wave fieldDirectional focused wave fieldConclusionComparison between numerical simulations and experimentsAnalytical methodsWavemaker controlNumerical checkingsIterative corrections1-Phase lag refinementIterative corrections2-Amplitude and phaseIterative Corrections